[cfe-commits] r67823 - /cfe/trunk/lib/Sema/SemaType.cpp

[Chris Lattner]

On Mar 27, 2009, at 9:56 AM, Douglas Gregor wrote:

>> Does this seem reasonable?
>
>
> Yeah, I thought about this after I'd committed the current version.  
> It's a heck of a lot of work/code for one error message, especially  
> if we get to the point where we're doing overload resolution to  
> determine whether there is a function that can take zero  
> arguments... but it's the right thing to do.

The fixit doesn't have to be perfect, just default to *not emitting  
the fixit*.  It would be perfectly fine to only emit the fixit in the  
simple case when it's a function that returns a pointer to struct, and  
ignore the overloaded-> case etc.

We'd miss the opportunity to offer fixit advice, but it is better than  
not offering advice for the "C case".

-Chris